University of Michigan Solar Car Team aimed to achieve their best-ever finish at the Bridgestone World Solar Challenge in Australia, in October 2017. It meant introducing new technologies and features to make their race vehicle, Novum. That included relying on Thuraya satphones for mission-critical data acquisition, machine learning, and team communications – and fitting Novum with a similar type of solar cells used on the panels of satellites.
Machine-learning via Thuraya satellite helps the University of Michigan take runners-up spot at world solar event.
As a long-term UoM team sponsor, Thuraya was delighted to supply six IP Voyager data terminals, six Thuraya XT-PRO satphones, and unlimited airtime for the race. The broadband links were vital to a critical aspect of the UoM race strategy – the M2M acquisition and synthesis of real-time weather data. Since 2015, UoM had downloaded machine-learned weather data directly from the servers of sponsor IBM. However, for the 2017 event, head strategist Alan Li decided to create a bespoke machine-learning model to handle weather data. This would allow the UoM team to acquire and synthesize data from a wider range of weather forecasters during the race and give the team better control of parameters. Alan says: “IBM agreed we could use its PAIRS application-programming interface to develop our own machine-learning model. We wrote the code from scratch and it took about six months to build, train and validate the model.”
